Magical Science

Arthur C. Clarke famously suggested that any sufficiently advanced technology would be indistinguishable from magic. This suggests another maxim: any insufficiency developed philosophy of science is incapable of distinguishing between science and magic.

We all have our own philosophy of science, our conceptual framework for understanding scientific topics. In the best case, our personal philosophy of science informs us of the limitations of scientific knowledge, allows us to put research into a wider context, and ensures we remember that the work of the sciences is still at heart an entirely human endeavour. Alas, few of us have such a clear view of the sciences. Far more widespread is a kind of pervasive mythos we might call ‘magical science’, which affords to the image of science unlimited future power, and to scientists an awesome capacity to divine the truth through singular experiments, like a Roman haruspex reading animal entrails to predict the future.

Magical science has the dubious honour of being the only superstition widely encouraged today. We are all too frequently adamant that science has all the answers, science is the royal road to truth, that we can trust in the science... I notice that even the British Prime Minister has taken to invoking magical science in his speeches these days to validate his increasingly dubious actions. At heart, magical science may seem harmless, a mere rose-tinted vision of the work of scientists, one that tries to account for all the successes of our various research networks without any attempt at balance or insight. We typically overlook this kind of naive enthusiasm for scientific achievement on the basis that it's at least ‘supporting the right team’. Yet it becomes increasingly clear that blind support for science can manifest in ugly ways, even in ways that can prevent the sciences from working, plunging research into the debilitating condition of pseudoscience, as previously discussed.

The perceived infallibility of the sciences as truth-seeking procedures clashes worryingly with the necessity of scientists making mistakes, and thus magical science leads to anger at scientists when the actual scientific work is not as wondrous as it is imagined it should be (as with the ugly 2009 L'Aquila trial, where terrible earthquakes in Italy were not successfully predicted and the scientists blamed), or when any scientist speaks out against a claim that has been proclaimed unshakably true by its advocates. It is precisely because magical science is incapable of distinguishing science from magic that it represents a far greater danger to scientific endeavours than other philosophies, perhaps even so-called ‘anti-science’ philosophies. What deceives us here, what elevates scientists to their misguided role as flawless augurs rather than researchers struggling with ambiguous data, are the bad habits we have learned from the manifestations of science in fiction, where magical science is the norm. If we wish to see the work of the sciences with clearer eyes, we may have to start by putting some of the most iconic characters in fiction on philosophical trial.

Sherlock Holmes and the Flawless Investigation

It is sometimes remarked that in creating Sherlock Holmes, Sir Arthur Conan Doyle produced the first hero of ‘the scientific age’. The Victorians were the ones who coined the term ‘scientist’ and it was their obsession with the sciences that set the scene for the unfolding technological transformation of the world over the next century and a half. We tend to treat the character of Holmes as significant mainly for crime fiction, as the archetype from which all whodunits descend - but Holmes, quite unlike a Raymond Chandler or Agatha Christie detective, is always a practitioner of magical science. Partly, this proceeds from the inherent parsimony of storytelling whereby all questions will eventually be answered because everything is there by the author’s design. Partly, however, it proceeds from Holmes’ essential power - which upon closer inspection is not deductive reasoning at all, but rather the infinite convenience possible solely in literature.

Doyle gives Holmes a quite impossible access to every conceivable fact as a starting point, such that a berry stain or the smell of a particular tobacco can certainly be identified, and then (to pile on the absurdity) Holmes by purest chance always encounters a set of circumstances that allow for only one viable interpretation. This particular brand of tobacco, for instance, is sold in exactly one place in London... We thus end up admiring Holmes purportedly scientific form of investigation while what we ought to admire is the way Doyle effortlessly conceals the magical science entailed in this depiction by making it seem as if all of Sherlock’s deductions (and inductions) were strictly logical. Doyle has contrived a set of circumstances that Holmes, with his unlimited catalogue of facts, can be certain to solve. This makes Holmes a disastrous role model for scientists (or indeed, detectives!) since it is only through the meticulous construction of literary contrivance that he possesses any investigative power at all. This becomes clearest when Holmes relies upon facts we know are false - such as the ludicrous snake plot device in The Speckled Band, which entails behaviour implausible to coax out of any reptile. Holmes’ claims to be a man of science are rather fraudulent behind the scenes: he is simply the locus of a mythic depiction of magical science.

Neither is Holmes the only such character. Both Spock and Data in the worlds of Star Trek share this power of magical science - also manifested in these shows by the tricorder, which like Holmes spits out every required fact on demand and without error. Or consider Doctor Who from the third Doctor onwards: anything necessary is certainly known by the Time Lord, except when the story requires a convenient (and often temporary) amnesia for dramatic effect. That both Data and the Doctor had a spin at being Baker Street’s most eligible bachelor is not accidental, nor perhaps is Stephen Moffat’s concurrent time as showrunner for both Doctor Who and Sherlock... Magical science heroes seem to reaffirm our faith in the power of scientific knowledge, while also playfully exposing the quirky personalities of scientists. House, The Big Bang Theory, and much more besides all participate in a literary tradition that stems from the Sherlock Holmes tales, and is now seemingly dominated by his science fiction proteges. 

Yet these are not scientific heroes, but magical science heroes. They have exactly the facts and the circumstances to answer perfectly every time, without ever having to confront the ambiguity, indeterminacy, and incompleteness of an authentic scientific problem. They are to science what Superman is to police officers: naively idealized caricatures. They find the answers solely because they live in stories where uncovering the truth is possible by design. This is a wildly misleading template for scientific truth, and although we know these are ‘just’ stories, we somehow import our wilder beliefs about the sciences into our everyday thinking unless we are extremely careful. If we are to break this spell, we need a philosophy capable of distinguishing science and magic - and for this, we need a clearer understanding of ‘scientific truth’.

Desperately Seeking Truth

Even if we start with the acknowledgement that the sciences are capable of discovering or affirming truth, the question of what might qualify as a ‘scientific truth’ is far trickier than it seems. As the preceding discussion on pseudoscience made clear, we cannot simply append ‘scientific’ to known truths without distorting the essential ambiguities of the research process where we cannot in practice know if the apparent truth of a researched claim will hold in the future. In fact, we have a choice. We could align ‘scientific truth’ with the unshakeable deep truth of reality and thus admit that the claims asserted by scientists cannot be known as truth at all (effectively contracting the domain of scientific truth to concluded research programmes like optics). Or else we can align scientific truth with the body of beliefs held by scientists, with the inevitable consequence that such truths can be later revealed as false - or even abominable. We don’t even have to go back a century to find all manner of racist, sexist nonsense asserted as truth by those who identified as scientists.

Now those who buy into magical science have an easier job here, but only by being wildly dishonest about both truth and scientific methods. According to magical science, scientists uncover truth infallibly so all claims asserted by scientists are scientific truth. Thus if and when the circumstances shift we can ‘debunk’ or ‘discredit’ those responsible and say they were not really scientists at all, or even exclude their claims from consideration in the first place! This is where ‘pseudoscience’ has been used as a label, although as I have argued previously it is not a terribly viable way of using the term. Babette Babich has made even stronger - and oft misunderstood - claims about the way the discrediting associated with the term ‘pseudoscience’ serves as a dogmatic attempt to demarcate legitimate science, while all too frequently preventing any scientific enquiry from even beginning. Thus when this particular word comes out, it narrows scientific knowledge by declaring certain topics forbidden and out of bounds - and woe betide the researcher who goes on to try to report experimental results from such verboten fields...

The highly problematic implication of every attempt to discredit and thus demarcate ‘science’ from ‘pseudoscience’ must be that we cannot know when scientists assert a claim whether it will later need to be ‘debunked’. Thus faith in magical science is inevitably a distortion of the truth - for things we will say are scientific truths on this philosophy may later be ‘discredited’, or even discredited before they are considered at all. The alleged truths of magical science are thus only defended by ignoring the inevitable consequences of the inherent revisionism of scientific practice and pretending that the current consensus among researchers is ‘more true’ than it was yesterday and thus that now (and by implication, only now) we can trust everything scientists say as long as we are standing guard for those pernicious pseudoscientists who ruin it for everyone. To say that this is dangerous nonsense is easy; to replace it with a more sound philosophy of science will be much harder.

There might be a way out of this maze, but it would require us to think differently about the relationship between truth and the sciences. Part of what deceives us here is our desire to understand the truth in terms of a set of valid statements. Since we can point to scientific concepts we abandoned, like phlogiston (which was a hypothetical substance that made combustion possible), we want to assert a gradual improvement in the accuracy or scope of our ‘book of facts’. “We would not be fooled by phlogiston today,” we might think. Yet phlogiston was an important - and arguably entirely scientific - proposal that was merely discarded when our understanding of chemistry shifted such that combustion could be thought of in terms of a chemical reaction with oxygen.

The brutal truth of the ‘book of facts’ is that such a collection of statements today would theoretically contain far more ultimately false claims than it would in the 1770s, simply because the number of scientists and the diversity of research fields has increased dramatically we are now paradoxically more wrong than researchers in the 18th century (in terms of sheer numbers of errors made) - the inescapable consequence of asking both more and more difficult questions. What makes it feel as if we are now more right is knowing that phlogiston was to become replaced by a new understanding of chemical reactions and thus combustion and so forth. But this is largely an illusion caused by examining successful research programmes in hindsight.

Similarly, when I say phlogiston was ‘scientific’, I am projecting with hindsight since the term ‘scientist’ was not coined until 1834... researchers in the 1770s would not have described anything they were doing as ‘scientific’ - it is our desire to paint the sciences as something with a history of more than two centuries that makes us ‘claim’ both phlogiston and oxygen (not to mention Copernicus, Galileo, Newton and so forth) as part of the story of ‘science’, rather than the natural philosophy that those involved would have stated they were pursuing. Thus our ‘book of facts’ not only contains more errors than our predecessors two and a half centuries ago, it is not even entirely honest about its relationship with its own past. Add to this the unavoidable truth that this imagined ‘book of facts’ does not exist (for all that encyclopedias and their successors have wished to fulfil this role) and it begins to feel uncomfortably like we are deceiving ourselves - as if we have all fallen for the seductive confusions of magical science.

Legitimate Practices

We want to defend our intuitive impression of the sciences as truth-seeking, and also (in some nebulous sense) successful at doing so. How do we do it?

One option we can consider is that which I proposed in Wikipedia Knows Nothing: to switch our focus from facts (true statements) to practices (skills and equipment). To know how to use something - a polymerase chain reaction, an interferometer, a fractional distillator - is more a matter of knowing what to do than it is a ‘book of facts’, even though that knowledge also produces facts related to the equipment used (and any theories deployed to give a context to the reading of the instruments). Thus an astronomer armed with geometric theorems can use an interferometer to measure the diameter of stars, while an engineer can use an interferometer and the wave theories of light to measure very small objects precisely. The practices associated with both the equipment (the interferometer) and the theories associated with each specific usage give rise to facts - in this case, distances. The difference lies in what legitimizes the activity in question: on the usual conception of knowledge, if you had the facts you had legitimate knowledge if those facts were true and the reasons for justifying them were correct - which actually provides no means of knowing what is or is not legitimate since our criteria for legitimacy requires an appeal to something beyond the situation (the truth) that we cannot access directly. Conversely, when we view knowledge as a practice, what makes the facts legitimate is that we are using the tools correctly. In this context, we have recourse to everyone with the relevant knowledge of the tools entailed to verify the legitimacy of the practices used and hence the facts reported.

On this understanding of knowledge, unlike an appeal to the truth, we can construct a viable understanding of ‘scientific truth’, since certain equipment, certain theories can be uncontroversially attributed to the sciences, and their correct usage can be judged by anyone else with access to the same knowledge practices. On this path we can therefore distinguish between scientific truth (facts emerging from legitimate research practices) and errors, provided we allow the disagreements to be properly explored in any given research community. However, as Babich warns, this cannot happen if we rush in with a dogmatic cry of ‘pseudoscience’, since every attempt to discredit something a priori entails an outright refusal to think about a given topic at all. Ironically, such attempts to discredit effectively cause an outbreak of the condition of pseudoscience, in my sense (a state of disrupted communication where scientific work can no longer be pursued), since whomsoever speaks this word with the intent to discredit (and thus ignore something) signals the very breakdown of legitimate scientific disagreement required to understand whatever is (not) being discussed.

The deeper problem we encounter when we look more clearly at how scientists discover or verify truths is that the claims that are asserted soon exceed simple assertions of facts. Once they do, it requires another set of knowledge practices to disentangle the relationships between facts and conclusions - and these are not strictly scientific at all, for all that scientists engage (unknowingly) in these kind of interpretative philosophical practices every time they assert anything but the most trivial of claims. Indeed, precisely the crisis of contemporary sciences is that their application is not a scientific practice, but a philosophical one - and Einstein’s generation may have been the last where scientists spanned these disciplines rather than retreating behind specializations that narrow, rather than widen, the scope of our collective understanding.

It is small wonder that we seem to have arrived in a “post-truth” world: the attempt to make the only acceptable truths those that flow from scientific endeavours renders a great many of the truths that matter impossible to adequately discuss, precisely because the important truths (those that pertain to what we ought to do, for instance) could never be scientific and thus cannot be established solely by an appeal to the facts. Yet we keep looking to scientists to give us a certainty that is not in any way available through scientific methods - and as the L'Aquila trial in Italy demonstrated, we will turn upon those who do not live up to our insanely unrealistic expectations and even accuse them of committing crimes when they, inevitably, make mistakes. But it is we that have failed, by falling for such an impoverished understanding of the complexity of scientific research as that of magical science.

Breaking the Spell

The needs of a narrative require magical science for the very same role as arcane magic - as a plot device limited solely by our imagination - and the two are (in more ways than we tend to acknowledge) equivalent, exactly as Clarke foreshadowed. The problem is, the actual work of the sciences, the global cybernetic collaboration of scientists that began under that name in the 1800s and continues today, is magical solely in its lustre and not in its details. Yes, the collective technological achievements facilitated by the work of countless scientists is now already indistinguishable from magic in a great many situations. But the work of scientists is not magic, and is certainly nothing like the magical science of a Sherlock Holmes fable. When we mistake the two, when we treat a human who conducts scientific work as someone wielding all the sorcery of magical science to know, automatically, everything that needs to be known, we are not supporting scientific truth-finding at all, but making it far, far harder, and in the worst cases, rendering it entirely impossible.

I will not say we must stop enjoying the fantasy of magical science in our stories - escapism is mostly harmless, after all, even if it is not entirely blameless - but is it not perhaps about time we stopped pretending that our scientists are superheroes with magical powers to determine truth? Scientific truths are extremely specific, and much narrower than we want them to be - they are at their most precise precisely when their claims are most limited. The heroism of actual researchers is of a patient, humble kind, that requires time and substantial disagreements to bring about. It is neither as spell-binding as Holmes’ contrived deductions, nor as charmingly detached from human fallibility as Data or Spock’s inhuman resourcefulness suggest. Neither has any living scientist access to the unquenchable moral certainty of the later incarnations of the iconic Time Lord to guide them either. These role models all imply a role that is impossible to bring to life: we should be careful not to buy too deeply into such implausible exemplars, without dismissing entirely the hopes and ideals that they embody.

Actual scientific practice is amazing, but it is neither miraculous nor supernatural. It is rather mundane in its details, which never entail perfectly prophetic experiments, and always require a great deal more arguing about the possible interpretations of the facts than literature has ever depicted. When we cannot distinguish science from magic, we obscure scientific truth and the immense and heroic efforts required to produce and understand it. We do all our scientists a disservice when we mistake them for sorceresses and wizards, and we entirely dishonour the scientific traditions when we censor or revile researchers for not living up to our hopelessly elevated expectations of their truth-discovering powers.

If we cannot distinguish science from magic, we need to either improve our philosophy of science or else remain silent on scientific topics. As Holmes remarks: the grand gift of silence makes Watson quite invaluable as a companion, for scientists, much like Holmes, often need us to pay close attention to their work and their disagreements, so that together we can eventually reveal true claims about our world. When we work to silence and discredit others we disagree with, rather than remaining silent so we might hear those disagreements we are denying, we have destroyed the very conditions for any kind of legitimate scientific investigation to occur. If we truly wish to be friends of the sciences, perhaps we too ought to know how to hold our tongue and try to listen to the quiet whispers of the truth when the game is afoot.

Comments always welcome, especially the polite ones!

Comments

[Paul Gestwicki]

Your consideration of procedural knowledge is interesting, although it seems your examples draw from natural sciences and engineering. Do you see any difficulty applying this to social sciences?

Here is where my mind goes with that question. If one wants to show that an educational game is effective, one commonly uses pre- and post-game tests together with statistical analysis. An honest scientist would ensure that the instrument is reliable and valid, although I have seen countless papers that skip this step: this is just bad scholarship, indicative again that "science" is done by fallible people. More importantly, though, I have never seen any evaluation acknowledge that, if the game is having measurable impact in the measured areas, then it almost certainly has other impacts that have not been measured. It was Koster's famous Theory of Fun that got me first thinking along these lines, that games can teach more than intended. The "magic" here seems to be that truth is limited to the thing that is measured, and this seems an inverse of the other problems you raise: that is, the social scientist's claims (in this case) are too small to be usefully true rather than too broad. For example, if a game teaches multiplication tables and also teaches that you are more important than your classmates, then I'm not sure we should focus on the multiplication tables.

Along similar lines, your essay makes me think of those whose "tools" are mathematics. A well-developed formal proof should be clear to any reader who knows the methods of mathematics. "Truth" in such cases is predicated on the axioms, not any other reality. I have heard cynical science faculty refer to the social sciences as having "physics envy", and your essay makes me wonder if the physical sciences suffer from "mathematics envy."

[Chris]

Dear Paul,
The question of what the proper instruments of the social sciences might be is, I contend, a very good one, and one that has not been asked enough! Firstly, I should say I am not one of these people who thinks that there is a hierarchy of the sciences with Physics at the top and Sociology/[Insert preferred 'Kick Me' Social Science here] at the bottom. Actually, I think we love physics because it was an easy research field and we're envious that so much could be settled in a research field so easily. When other fields try to be physics, they go horribly wrong, because no two research fields are entirely alike.

All that said, I will tend to draw from natural sciences and engineering because my academic background is, in order of appearance, Physics with Astrophysics BSc (unfinished, 2 years only), Computer Science BSc, AI/Cognitive Science MSc, and then PhD effectively in philosophy. I'm aware of a bias in my writing here... but it is not that I have not engaged with social science, so much as I think the social sciences have problems they're not good at admitting they have.

Here's a really good point you make:

...the social scientist's claims (in this case) are too small to be usefully true rather than too broad. For example, if a game teaches multiplication tables and also teaches that you are more important than your classmates, then I'm not sure we should focus on the multiplication tables.

That is such a great observation! But isn't the problem here that we so desperately want to have means of quantifying topics that are difficult to quantify. The key instrument of social science is statistical analysis. I think this instrument can do excellent work! But it is also the vaguest of tools... you can't hang that much on it. What the social sciences have as a truly excellent tool is observation. But there's a problem in saying this.

Here, as in many places today, we crave the apparent certainty of scientific methods without really acknowledging that certainty isn't something most scientific fields are going to get. It's certainly out of reach in the medical sciences, where good research practice means never presuming the matter is closed to further investigation.

Having worked on a learning game myself (won an award for it, actually... for what little that means!), I'm mindful of the pressure being placed on games with designed purposes to prove they 'work'. I think that sense of a demand for proof is problematic - doubly so because it is extended into every corner of activity. The demand for proof that teaching is effective, for instance... that swiftly turns ugly, especially when that desire comes bouncing back in the form of demands to teach a certain way because that's the right way or the best way, or whatever justification is used. But learning is a foundation of animal experience all the way through to the elderly, who gradually lose that capacity (my objection to Koster's excellent book in a nutshell!). But whatever test we devise are going to turn into a self-referential issue - like training students to pass tests instead of teaching them skills. The problem here is that we don't trust teachers to teach (if we did, we wouldn't spend so much time trying to evaluate teaching - much less trying to enforce specific methods).

So where is the truth of the Social Sciences? In much the same place as the truths of the humanities. Not in the eye of the beholder (although without a beholder there can be no experience of truth) but in the veracity of the claims, in Jacques Rancière's sense.

Truth doesn’t bring people together at all. It is not given to us. It exists independently from us and does not submit to our piecemeal sentences. “Truth exists by itself; it is that which is and not that which is said. Saying depends on man, but the truth does not.” But for all that, truth is not foreign to us, and we are not exiled from its country. The experience of veracity attaches us to its absent center; it makes us circle around its foyer. First of all, we can see and indicate truths.

The thing is, we seem to desire all sciences to match up to physics because what we want is certainty, not truth. Truth is far more chimerical, far harder to design experiments for, than certainty, which can be won easily - although usually only dishonestly, as this piece expressly tries to undermine. The idea that a scientist (of any kind) has magical powers to reveal certainty, rather than the gift and the curse to investigate and wrestle with the problems is so damaging, and this is what hurts the social sciences. We want metrics so we can use them to win certainty. But the sciences are about trying to find truth against a backdrop of immense uncertainty - and never more so than when we humans ourselves are our subjects!

I do think the physical sciences have mathematics envy. This is a wonderful way of putting it! There's a desire to get to truths without acknowledging the utter dependence upon the axioms... well, you can get there, but what you can't get is the certainty that comes with the logic underlying maths. Mathematics has truth and certainty. Physics has truths that are expressible in mathematics. That's not the same. And it is indeed a source of unspoken envy.

So I suppose the question is: are any of the instruments of the social sciences good enough to reveal truth? Well, no. Not by themselves. But the statistical instruments are still instruments, they can still measure, they can still hint at the truth that, like an unseen moon that perturbs the orbit of its planet, can be detected but not seen. They have to rely on the same kind of hermeneutic reasoning that works for the humanities. The mistake we've made, I would say, is in thinking that now the sciences are so widespread we don't need the hermeneutics. And that's nonsense! The early twentieth century physicists (Einstein is the subject of the next philosophy of science piece in two weeks) were able to master the problems of physics because they understood the hermeneutics of physics. It's something nobody says. It's something I can't prove. But still... there is a moon circling this planet and I can feel its perturbations. Can you...?

Thank you very much for this thoughtful comment,

Chris.

[Sean Payne]

Very well thought out and written, does bring to mind the set by Dara Ó Briain where he says "science knows it doesn't know everything, otherwise it would stop...herbal medicines have been around for thousands of years, indeed it has, and then we tested it all and the stuff that worked became medicine!".

I know House got a mention here and it strikes me that he's a pretty bad doctor as, unlike a lot of other magical science superheroes, he makes a lot of mistakes before he nails the problem. Ironically maybe we perceive him as a flawed doctor simply because he does get it wrong a lot before he gets it right, in that he's not Holmesian in his accuracy?

On that point, are there any fictional magical scientists you would consider to be more accurate to the cause of pointing out that the scientific process isn't a series of perfect binary outcomes, or there aren't any as that negates the point of fantasy in the first place? Genuinely interested as my reading habits tend to be quite limited.

[Chris]

Hi Sean,
The comparison between House and Holmes is an interesting one... both are characters portrayed with weaknesses of personality (Holmes has his opium habit, for instance), but you're absolutely right - the formula for House was built around repeated mistakes and only eventually hitting the right answer. This is definitely a step on from Holmes in some respect... but it's also still magical science. The right answer is inevitably going to be found. The drama is in the process; at a time when the police procedural was flourishing, the team behind House found a way to make a medical procedural, and casting Hugh Laurie was the icing on the cake. But I think the character of House still has Holmes power of 'all the facts at hand'... it is rather that the structure of the story exposes House to failure before reaching success, whereas Holmes simply ticks along in his own head until the bell chimes.

I have found very few portrayals of scientific work in fiction that grapples with the ambiguity of genuine research situations, but one that definitely gestures in this direction is Walter Jon Williams' Days of Atonement - the trouble with recommending this book is that Williams has so many other, better stories to his name (Angel Station, Aristoi, and Voice of the Whirlwind to name just three) it makes this one hard to recommend. (It also tends to sit badly with non-religious readers, although the role of religion in the story is only to put the characters into a context).

Many thanks for your comment!

Chris.

[Joaquín Ágreda]

Great post, Chris. As a sociologist turned market researcher, I couldn't agree more.  Although I tend to be humble and open to discussion while offering my findings on a given research, when a client shows some unease that I feel is going to be too tedious to dispel, I just have to say "algorithm" and that kills any discomfort at once. A (magic) trick of the trade.

[Chris]

Hi Joaquín,
It is sometimes quite surprising how a suitably chosen word, or a chart, carries 'the smell of science'... ;) Although the thoughts expressed here evolved out of recent concerns I've been having about the costs of turning a blind eye to poor philosophy of science, I rather suspect I began on this path when I realised that my consultancy's early attempts at player modelling were being taken more seriously because they 'smelled' scientific.

Many thanks for your kind words!

Chris.